I've read the comments from Guido and Martin, but they don't convince me
in changing my -1.

As you say: it's difficult to get support for optimizations such a
slicing and concatenation into the core. And that's exactly why I want
to keep the door open for developers to add such support via extension
modules. With the 8-bit string implementation this was never possible. 

With the Unicode implementation and the subclassing support for builtin
types, it is possible to add support without having to go through all
the hoops and discussions on python-dev or py3k lists.

I'm also for making Python faster, but not if it limits future
optimizations and produces dead-ends. The fact that such optimizations
haven't been implemented yet doesn't really mean anything either -
people are not yet using Unicode intensively enough to let the need arise.